In recent years, technologies for recommendation systems, search, and advertising have been developing rapidly. The goal of our team is to search for, conceive, implement, and deploy new technologies. We collaborate with various product services within Yandex: Music, KinoPoisk, Market, Advertising, Search. An example of a technology we are actively developing is personalization using transformers. But it doesn't end with transformers — we also develop other technologies, including neural network ranking, graph neural networks, reinforcement learning for recommendations, and we are constantly searching for new ones.

We have real R&D:

• Development of key core technologies to improve the quality of recommendations, search, and advertising across the entire Yandex ecosystem
• Research into potential technologies, studying SOTA approaches used in other companies
• Active interaction with Yandex's largest services, used by millions of people

What tasks await you

• Reading articles and searching for new technologies
• Engaging in the full R&D cycle: preparing data, training models, evaluating their quality and interpreting results, participating in the deployment of technology, and creating convenient tools for its integration across the entire ecosystem
• Participating in the development of our frameworks for working with data, training, and deploying models

We expect that you

• Have a strong algorithmic background
• Are proficient in Python
• Possess excellent mathematical preparation, especially in the context of machine learning
• Have a good knowledge of the basic theory of deep learning, are on familiar terms with transformers
• Have worked with neural network models and distributed training

It will be a plus if you

• Have written your own wrapper framework for distributed training of neural network models
• Have deployed neural networks to production
• Have engaged in representation learning on large volumes of data, such as graph-based, text, and image models
• Have experience in low-level language (C++) development, profiling of neural network training and inference
• Have prepared publications for top conferences (RecSys, CIKM, NeurIPS, SIGIR, WWW)
• Have developed recommendation systems
• Have worked with large volumes of data, understand the MapReduce paradigm and basic SQL well

Working conditions

• Official employment
• Work in a team of talented experts from whom you can learn a lot
• A culture of openness and mutual assistance
• The opportunity to quickly see the results of your work and create services used by millions of people
• High total income, bonuses every six months
• Flexible schedule: we don't control who comes and goes at what time, the main thing is to complete tasks
• An extended voluntary health insurance program including dentistry, check-ups, etc.
• The opportunity to participate in Yandex's educational programs, lectures, and meetups